U.S. to decide soon on GM's request to deploy cars without steering wheels

L4sBot@lemmy.worldmod to Technology@lemmy.world – 219 points –
U.S. to decide soon on GM's request to deploy cars without steering wheels - Autoblog
autoblog.com

U.S. to decide soon on GM's request to deploy cars without steering wheels::U.S. regulators will soon decide on a petition filed by General Motors' Cruise self-driving technology unit seeking permission to deploy up to 2,500 self-driving vehicles annually without human controls, a top auto safety official said on Wednesday.

98

I don't understand how cars would work on their own with no input from a driver small-scale

I understand being able to type in "Drive to Walmart" and it can back out of your driveway and go to Walmart, but then what? It goes into the parking lot? It finds the first available space? What if you just wanted to go there to pick something up curbside? How can you tell it to go to a specific stall? What if you're disabled and need to go to the handicap space? How can it tell if your authorized to use that space?

There's so many little nuances that I don't understand not being able to have a steering wheel to take control of and manually do things at some point.

Read the article, this isn't talking about consumer-owned vehicles but the Cruise Origin robo-taxi service. They're small autonomous shuttle-style cars.

Basically GM reinvented the bus but made it smaller.

So to answer "How does it park at Walmart" - it takes the passenger to the front and drops them off then continues on its way. I believe the intent/current trials using Bolts have an app similar to Uber, you put in your current location + destination, then it comes and gets you, then drops you off.

Almost 0 value in removing a steering wheel or any kind of input to a consumer-owned car like that, makes some amount of sense for robo-taxis. (They specifically wanted passengers sharing the ride to face eachother to ease safety concerns, and they probably don't want random Joe getting up at the emergency controls and driving it off road)

9 more...

Back in the 80s or 90s GM (specifically Buick) teased a car with no steering wheel. It instead used joysticks. I’m curious if GM is basically thinking of that. Something more motor friendly, but joysticks also free up space for either more electronics (bad idea) or more safety equipment. The other thing people forget about is that a steering wheel is a giant spear aimed at drivers in a collision. We’ve gotten better about breakaway systems and shears, but it’s another point of injury and failure. The more enclosed a cabin the better. Anyways, all this to say that it might be that direction that GM is thinking and not a fully no input vehicle. It could also be a fleet based vehicle that only drives on main roads which effectively makes it a train that follows a “digital track” and doesn’t allow for nuance and is built for taxi service.

You just made me realize that we created a disconnect between the driver input and the car response on most thing except for the steering that for whatever reason is still a physical column down to the direction.

At this point electronic joystick and steerings are ancient in the PC gaming space, I don't see why that physical link is still required.

Infiniti on the Q50 released the first “direct adaptive steering” which was fancy marketing for steer by wire where no column was supposed to be present. This made it so consumers still had the same feeling, but it allowed for cool things like not having a rod aimed at your body, closed up another point of egress into the cabin for critters and water, and also gave you the ability to have it account for road undulations and wind so if you held the wheel straight even on a windy day it would adapt and steer straight. People however freaked out about steering not having a physical link and so Infiniti added in a column that would effectively reconnect if for some reason the steering ever stopped working. But it ruined the idea behind it. Anyways, consumers are kind of what holds us back. We all think of things having to be done a certain way without realizing there could be better ways of doing something. Side mounted joysticks, like a plane, would allow for people without legs for instance to drive cars. People with fine motor skills could be more precise and software could account for a shake in their hands.

A few companies are starting to experiment with brake by wire and throttle pedals haven’t been physically cable linked for decades at this point. Why do we still have steering wheels like that?

Not saying alternatives to steering are a bad thing, but there is also an issue of feedback and customer expectations. People like what they know/are used to. That’s why EVs had to add a lurch option and additional sounds. It throws people off mentally when part of the standard experience is missing.

Joysticks in theory would be an improvement, but let’s be honest you’d basically have to retrain people on how to drive it. Just a person gets additional training even to drive a forklift. And let’s be honest even if mandated not everyone would, and there would be wrecks. Not counting because of the learning curve it’d sell less, and it’d get bad press for every wreck.

I suspect the general consumer would be willing to hand control over fully, than have to spend extra to relearn how to drive their vehicle. We’ve been trained that self driving cars are the future for multiple decades now.

Because the opportunity and severity of failure of the physical steering wheel is an order of magnitude or more greater than any other system that has been replaced by electronic systems in most cars.

I would argue that steerings are already fully electronically controlled in self-driving cars or have been partially controlled for a while now in traction control systems.

Putting cars aside, most large aircrafts are fly-by-wire and are really reliable in that regard.

We trust our lives to wires all the time, brakes, acceleration, airplane controls, elevators. This is no different, you just put in enough redundancy to make failures safe

I've got an idea, how about we put that sort of control on a submarine?

I think some sort of joystick would be a great solution. Maybe over 15mph it gets disabled and autopilot takes over to take you to your destination (i.e. Walmart parking lot) and then when it slows down the joystick can be used to direct the car to where you want to park.

I can see a joystick becoming dangerous at high speeds though, which is maybe why they have stuck to steering wheels.

Joysticks don't take much to accidentally push forward or ya k back on, suddenly speeding up or stopping your vehicle, or if a kid or a pet starts misbehaving and knocks it to the side.

I think a rotary knob will be more intuitive than a joystick. Input is fuctionally the same as a steering wheel, and more likely to require less specialized training and adjustment.

With a joystick you can get rid of the gas and brake pedals though, with forward and backward movement. A knob would require the pedals to remain I imagine.

Of course. But I think id much prefer steering and acceleration to be decoupled

Exactly. Even if this car was 100% fully capable of self-driving, I would want the wheel to do the little final inputs like you describe. I can talk to my computer and have it do things, but I haven't ditched my keyboard.

9 more...

No, no, no, no. This is right up there with my state wanting to let 18-year-olds carry concealed deadly weapons in classrooms with no permit. A deal breaker for my continued participation in this society.

I recently learned that the state where I’m from considers a baton a deadly weapon so it can’t be purchased as a means of defense, but an Ak-47 is perfectly fine to purchase AND open carry.

Great gods. I guess the AK is considered a hunting rifle or a paperweight?

I'm tempted to ask if you're from the South, since this is exactly the garbage that too much of the region would champion. But honestly, the upper Midwest seems just as bad when it comes to "Muh guns!"

To be fair I'd have a hard time concealing my ak47. It doesn't get fired much, ammo is expensive. Most of my ammo is for shotties and the .22 for putting down nuisance and injured animals.

Yeah self driving cars are totally the same thing as giving a teenager a gun in a school.

Given their track record, yes.

Their track record is literally better than the track record of the average human driver. The data doesn't back up what you're saying at all

Source?

https://storage.googleapis.com/waymo-uploads/files/documents/safety/Safety%20Performance%20of%20Waymo%20RO%20at%201M%20miles.pdf

This is waymo, not cruise, but it's a comparison between 1 million miles with no humans behind the wheel and the human average. That's about 80 years worth of human driving, and while that's not long enough to provide meaningful data on reducing fatalities, they do show that self driving cars reduce both the frequency and severity of more minor accidents

2 more...
2 more...
5 more...
5 more...
5 more...

I wonder if this came from engineering or the marketing department.

I'm getting "Let's do this, and if it fails (which it will), it'll look like we're really confident in our self-drive and are a challenger in the market" vibes.

Even if you have excellent self-drive, there is no logical reason not to have a backup steering wheel just to intervene in case. Tbh, I had no idea they were even in the self-drive market which may be their true problem. No one really knows.

I remember being shocked that my third generation smart phone didn't have a pull out keyboard. Or that headphone jacks became a casualty. I think in the long run this is the goal.

Next step, you can't use the car you paid for without a subscription for the self driving service. Want to drive yourself instead to save money? Nope, you bought a car without a steering wheel.

You won't even be able to buy these cars, they'll just be subscription only. If you're not paying the subscription, it won't even show up

if there’s no one in the driver’s seat to pay attention, then why would you have a wheel to intervene “just in case”?

Has anyone proven a technology for that long and that consistently that it is safe. I've seen quite the contrary and sensible legislators expect a human there in case of issue.I'd expect that for years before a reasonable level ofconfidentce is reached.

Well you know now! So it worked.

Funny thing about ai is there isn't really a moat. Once an idea is out there it's easy to catch up, so don't be surprised to see the big players ahead now get caught up by competitors very quickly.

Self driving is a lot more than just AI though. Our current AI gets really smart really quickly, but fails way too often to be used in critical roles like driving, which is why most of the code in these self driving cars isn't actually AI. They use it to predict the path of other cars, decipher data, and make high level decisions, but the actual control of the car (steering, brakes, etc) is all traditional programming. Waymo even talked about how they used to have more AI, and removed some in favor of traditional logic

From the article, the cars will be used as autonomous taxies. It is too bad. I was hoping that by removing the steering wheel, GM would have to add in the most important important self driving car feature: a release of liability. I will not fully trust autonomous car controls until I stop being legally responsible for the actions of my car.

Public transport should be the primary target for the technology at this time. At least in parts of Europe, the primary issue with increasing public transport capacity is lack of drivers.

I fantasize about the day that I could get into a self driving car and just shout a destination at it, then go there. My commute is over an hour a day and I could use that time for myself. I don't think that day will be here soon.

These companies need to demonstrate that their vehicles are capable (more capable than MOST drivers), and they need to do it transparently. Whats more important to me, would be the manufacturer taking the liability, licensing, and most importantly take responsibility for their mistakes (again openly).

I do as well, but then I remember the insane amount of money and man power it takes to keep things working well (like airplanes, subs, etc) and realize exactly zero people would realistically do that. So it's great in theory, but really, really far off from being something I'd trust.

For years I argued with people that with some small changes to our infrastructure and some regulation, this would be something we could accomplish in just a few short years. I still feel that way somewhat, but every time a bridge collapses, or I hit the same pothole I've hit for months it erodes.

That dream is a reality right now in at least a couple of cities. At this point its a matter of expansion and scaling, because right now it's geofenced into a couple of small areas

I'm pretty sure the technology is not ready to go this far

It’s not. It absolutely is not

It already exists... https://youtu.be/ilcrMz_hWz8

No, actually it doesn’t.

You’ll notice that a) Waymo cars kill people (there’s pending litigation,) and b) they’re only in places with basically perfect weather.

Self driving cars cannot handle any sort of real winter driving.

I can't find anything about waymov cars killing people. Not saying it didn't happen, but id like to see your sources on some (assuming more than one since it's plural) of the deaths.

Regardless, the technologies existence is what is being debated here. Not the moved goalposts of "operable in all weather" or "has never killed someone" using those same goalposts I could claim that passenger aircraft isn't here yet since you can't take off in bad weather & they have killed people

apologies... I seem to have conflated a few items from my feed. There were no deaths. that said, Waymo is defending itself from a lawsuit in SF (iirc) because of an uptick in accidents, and the NTSFB peeps say that waymo is involved in the most accidents (per road hour, IIRC) of all the autonomous vehicles.

They're also suing (or were suing) the California DMV to keep their accident records secret as a "trade secret". lol.

That's very interesting, actually. Because manufacturers of self-driving cars say they're only drive-assisted, not autonomous. To dodge liability for their cars causing accidents. If the car is without a steering wheel and therefore fully autonomous, and the car runs somebody over, is the car gonna go to jail?

Yeah don’t know how well that works in practice. Either I’m engaged and driving or I’m not. Letting the car drive while I continue to pay attention and somehow anticipate with superior reaction time that the car is going to make a mistake seems like more stress than it is worth. My current car has lane keep assist, which is helpful, but if I’m going to give full control of the car to the computer, then I don’t think I can/want to be on standby. Seems like more stress than it is worth.

The cars they're talking about here are fully autonomous, not driver assisted. It's a completely different stack from their commercial vehicles that are driver assisted, and they're not for sale yet.

If you run somebody over, unless you did it deliberately or with criminal negligence, you aren’t going to jail.

From the article, this looks like it's for GM's "Cruise" program, which is already out there in limited scope in a couple cities. It's aself-driving car service limited to a small area of San Francisco and.. I want to say Austin?

They're already operating vehicles that are essentially "self-driving" now. This is about rolling out a new class of vehicle using the same technology, but without the human controls.

I don't know a lot about the service, or what, exactly it does, but I suspect it works well because the area the vehicles operate in is extremely limited and the vehicles can have an incredibly detailed, and up-to-date map of that area. I'd also wager the area selected is free of most obstacles and has only one type of terrain, i.e., "downtown low-speed streets" or similar.

That said, I can't imagine the NTSHA will allow a vehicle on the road without any sort of manual emergency control mechanism in place. Though, it may be very rudimentary, like others have suggested, a joystick and a throttle/brake intended to get the vehicle somewhere safe so people can get out.

This is Level 4 automation, and even it needs a human override. Can't override without a steering wheel.

I think this whole plan is just a gimmick to dangle that carrot in front of the public, fooling them into thinking Level 5 full automation is just around the corner, when it really, really, really isn't.

The override system likely exists. It obviously exists now as remote operators can intervene on the current fleet without being present.

The human override is done by an employee in a control center, not by the passengers. They can and do override without using the steering wheel, and that's never going to change. The steering wheels are already obsolete, because passengers aren't even allowed to sit up front or touch the wheel.

This is as level 5 as we're ever going to get.

And what happens if something unexpected (e.g. badly signalized road works) comes up? With current implementations you can you take over. But what do you do without wheel?

Actual answer: waymo and GM both have control centers that can direct the car remotely. I know that in waymo's case they don't actually remote pilot the car, they just override the cars path finding and the car goes down the route the remote operator makes. I'd assume GM does something similar

Having just seen several Cruise vehicles being recovered by human drivers, this is quite early for such a request. I think the hype cycle is drying up for them and they need to keep pretending they're doing better than they really are.

Everyone keeps saying no in this thread, but I trust that it will be a better driver than my 80 year old grandpa who can't see 10 feet busy still manages to retain their drivers license.

It's okay to say "No" to two things at the same time without having to choose one or the other.

Except that in this case, one might prevent/reduce the other. There is a certain amount of mutual exclusivity here. We can't take drivers licenses away from old people, but we might be able to get them in an autonomous car.